FLAMES: Improving LLM Math Reasoning via a Fine-Grained Analysis of the Data Synthesis Pipeline

• URL: http://arxiv.org/abs/2508.16514v1
• Date: Fri, 22 Aug 2025 16:37:40 GMT
• Title: FLAMES: Improving LLM Math Reasoning via a Fine-Grained Analysis of the Data Synthesis Pipeline
• Authors: Parker Seegmiller, Kartik Mehta, Soumya Saha, Chenyang Tao, Shereen Oraby, Arpit Gupta, Tagyoung Chung, Mohit Bansal, Nanyun Peng,
• Abstract summary: We introduce FLAMES, a Framework for LLM Assessment of Math rEasoning Data Synthesis.<n>Our FLAMES experiments provide valuable insights about the optimal balance of difficulty and diversity of synthetic data.<n>We develop the FLAMES dataset, an effective blend of our novel and existing data synthesis strategies.
• Score: 71.19227942708741
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent works improving LLM math reasoning with synthetic data have used unique setups, making comparison of data synthesis strategies impractical. This leaves many unanswered questions about the roles of different factors in the synthetic data pipeline, such as the impact of filtering low-quality problems. To address this gap, we introduce FLAMES, a Framework for LLM Assessment of Math rEasoning Data Synthesis, and perform a systematic study of 10 existing data synthesis strategies and multiple other factors impacting the performance of synthetic math reasoning data. Our FLAMES experiments provide several valuable insights about the optimal balance of difficulty and diversity of synthetic data. First, data agents designed to increase problem complexity lead to best improvements on most math metrics. Second, with a fixed data generation budget, keeping higher problem coverage is more important than keeping only problems with reliable solutions. Third, GSM8K- and MATH-based synthetic data can lead to improvements on competition-level benchmarks, showcasing easy-to-hard generalization. Leveraging insights from our FLAMES experiments, we design two novel data synthesis strategies for improving out-of-domain generalization and robustness. Further, we develop the FLAMES dataset, an effective blend of our novel and existing data synthesis strategies, outperforming public datasets on OlympiadBench (+15.7), CollegeMath (+4.5), GSMPlus (+6.5), and MATH (+3.1). Fine-tuning Qwen2.5-Math-7B on the FLAMES dataset achieves 81.4% on MATH, surpassing larger Llama3 405B, GPT-4o and Claude 3.5 Sonnet.

Related papers

• Should I use Synthetic Data for That? An Analysis of the Suitability of Synthetic Data for Data Sharing and Augmentation [16.434161021014692]
  We study three prominent use cases: (1) Sharing synthetic data as a proxy for proprietary datasets to enable statistical analyses while protecting privacy, (2) Augmenting machine learning training sets with synthetic data to improve model performance, and (3) Augmenting datasets with synthetic data to reduce variance in statistical estimation.
  arXiv  Detail & Related papers  (2026-02-03T17:52:57Z)
• Understanding the Influence of Synthetic Data for Text Embedders [52.04771455432998]
  We first reproduce and publicly release the synthetic data proposed by Wang et al.<n>We critically examine where exactly synthetic data improves model generalization.<n>Our findings highlight the limitations of current synthetic data approaches for building general-purpose embedders.
  arXiv  Detail & Related papers  (2025-09-07T19:28:52Z)
• Synthesis by Design: Controlled Data Generation via Structural Guidance [7.938713951512933]
  We propose to extract structural information with generated problem-solving code from mathematical reasoning.<n>Our approach produces 39K problems with labeled intermediate steps and a 6.1K-problem benchmark of higher difficulty.<n>Results on our benchmark show that model performance declines as reasoning length increases.
  arXiv  Detail & Related papers  (2025-06-09T11:38:23Z)
• Scaling Laws of Synthetic Data for Language Models [132.67350443447611]
  We introduce SynthLLM, a scalable framework that transforms pre-training corpora into diverse, high-quality synthetic datasets.<n>Our approach achieves this by automatically extracting and recombining high-level concepts across multiple documents using a graph algorithm.
  arXiv  Detail & Related papers  (2025-03-25T11:07:12Z)
• Evaluating Language Models as Synthetic Data Generators [99.16334775127875]
  AgoraBench is a benchmark that provides standardized settings and metrics to evaluate LMs' data generation abilities.<n>Through synthesizing 1.26 million training instances using 6 LMs and training 99 student models, we uncover key insights about LMs' data generation capabilities.
  arXiv  Detail & Related papers  (2024-12-04T19:20:32Z)
• Unleashing LLM Reasoning Capability via Scalable Question Synthesis from Scratch [54.12139707822201]
  We propose ScaleQuest, a novel, scalable, and cost-effective data synthesis method.<n>By generating diverse questions from scratch, we produce a dataset of 1 million problem-solution pairs.<n>Our experiments demonstrate that models trained on our data outperform existing open-source datasets.
  arXiv  Detail & Related papers  (2024-10-24T12:42:04Z)
• OptiBench Meets ReSocratic: Measure and Improve LLMs for Optimization Modeling [62.19438812624467]
  Large language models (LLMs) have exhibited their problem-solving abilities in mathematical reasoning.<n>We propose OptiBench, a benchmark for End-to-end optimization problem-solving with human-readable inputs and outputs.
  arXiv  Detail & Related papers  (2024-07-13T13:27:57Z)
• Synthetic Oversampling: Theory and A Practical Approach Using LLMs to Address Data Imbalance [16.047084318753377]
  Imbalanced classification and spurious correlation are common challenges in data science and machine learning.<n>Recent advances have proposed leveraging the flexibility and generative capabilities of large language models to generate synthetic samples.<n>This article develops novel theoretical foundations to systematically study the roles of synthetic samples in addressing imbalanced classification and spurious correlation.
  arXiv  Detail & Related papers  (2024-06-05T21:24:26Z)
• JiuZhang3.0: Efficiently Improving Mathematical Reasoning by Training Small Data Synthesis Models [110.45794710162241]
  Existing work either collects large-scale math-related texts for pre-training, or relies on stronger LLMs to synthesize massive math problems. We propose an efficient way that trains a small LLM for math problem synthesis, to efficiently generate sufficient high-quality pre-training data. We leverage it to synthesize 6 million math problems for pre-training our JiuZhang3.0 model, which only needs to invoke GPT-4 API 9.3k times and pre-train on 4.6B data.
  arXiv  Detail & Related papers  (2024-05-23T09:43:19Z)
• EPIC: Effective Prompting for Imbalanced-Class Data Synthesis in Tabular Data Classification via Large Language Models [39.347666307218006]
  Large language models (LLMs) have demonstrated remarkable in-context learning capabilities across diverse applications.<n>We introduce EPIC, a novel approach that leverages balanced, grouped data samples and consistent formatting with unique variable mapping to guide LLMs in generating accurate synthetic data across all classes, even for imbalanced datasets.
  arXiv  Detail & Related papers  (2024-04-15T17:49:16Z)
• Key-Point-Driven Data Synthesis with its Enhancement on Mathematical Reasoning [110.80663974060624]
  Key-Point-Driven Data Synthesis (KPDDS) is a novel data synthesis framework that synthesizes question-answer pairs. KPDDS ensures the generation of novel questions with rigorous quality control and substantial scalability. We present KPMath, an extensive synthetic dataset tailored for mathematical reasoning, comprising over 800K question-answer pairs.
  arXiv  Detail & Related papers  (2024-03-04T18:58:30Z)
• ChatGPT Based Data Augmentation for Improved Parameter-Efficient Debiasing of LLMs [65.9625653425636]
  Large Language models (LLMs) exhibit harmful social biases. This work introduces a novel approach utilizing ChatGPT to generate synthetic training data.
  arXiv  Detail & Related papers  (2024-02-19T01:28:48Z)
• Generating Faithful Synthetic Data with Large Language Models: A Case Study in Computational Social Science [13.854807858791652]
  We tackle a pervasive problem in synthetic data generation: its generative distribution often differs from the distribution of real-world data researchers care about. We study three strategies to increase the faithfulness of synthetic data: grounding, filtering, and taxonomy-based generation. We conclude this paper with some recommendations on how to generate high(er)-fidelity synthetic data for specific tasks.
  arXiv  Detail & Related papers  (2023-05-24T11:27:59Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.